Real-time monitoring method, device and equipment for public opinion risk of live broadcast room and storage medium

Abstract

The embodiment of the invention relates to the field of public opinion monitoring, and discloses a live broadcast room public opinion risk real-time monitoring method and device, equipment and a storage medium, and the method comprises the steps: collecting the language information of a live broadcast room; the language information comprises anchor speaking information; identifying the language information, and determining text information corresponding to the language information; performing context analysis on the text information by utilizing a pre-constructed natural language model, and determining semantic content and / or emotional tendency; judging whether the language information has a public opinion risk or not according to semantic content and / or emotional tendency, and determining a public opinion risk level when judging that the public opinion risk exists; and when the public opinion risk level is higher than a preset threshold value, generating warning information for prompting an anchor based on the language information. According to the monitoring method disclosed by the invention, the problem of inaccurate public opinion risk identification is solved, so that the direct broadcasting room public opinion risk is accurately monitored, and intelligence and refinement of public opinion monitoring and response are realized.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of public opinion monitoring, and in particular to a method, device, equipment and storage medium for real-time monitoring of public opinion risks in a live broadcast room. Background Art

[0003] In the e-commerce live broadcast scenario, the real-time interaction between the anchor and the audience has become a key link in brand communication and product sales. However, the current public opinion monitoring system mainly relies on keyword matching technology, which is difficult to effectively identify complex language tendencies in the context, resulting in inaccurate identification of public opinion risks and untimely processing. Summary of the invention

[0004] The purpose of the present invention is to at least provide a method, device, equipment and storage medium for real-time monitoring of public opinion risks in a live broadcast room, which can at least solve the problem of inaccurate identification of public opinion risks, at least accurately monitor the public opinion risks in the live broadcast room, and realize the intelligence and refinement of public opinion monitoring and response.

[0005] To solve the above technical problems, at least one embodiment of the present application provides a method for real-time monitoring of public opinion risks in a live broadcast room, including:

[0006] Collecting language information of the live broadcast room; the language information includes the host's speech information;

[0007] Identifying the language information and determining text information corresponding to the language information;

[0008] Performing context analysis on the text information using a pre-built natural language model to determine the semantic content and / or sentiment tendency corresponding to the text information;

[0009] Determine whether the language information has a public opinion risk based on the semantic content and / or emotional tendency corresponding to the text information, and determine the public opinion risk level corresponding to the language information when it is determined that there is a public opinion risk;

[0010] When the public opinion risk level corresponding to the language information is higher than a preset threshold, a warning message for prompting the anchor is generated based on the language information.

[0011] At least one embodiment of the present application further provides a device for real-time monitoring of public opinion risks in a live broadcast room, comprising a data collection module for collecting language information of the live broadcast room; the language information includes the host's speech information;

[0012] A text extraction module, used to identify the language information and determine the text information corresponding to the language information;

[0013] An analysis module, used to perform context analysis on the text information using a pre-built natural language model to determine the semantic content and / or sentiment tendency corresponding to the text information;

[0014] A public opinion risk judgment module is used to judge whether the language information has a public opinion risk according to the semantic content and / or emotional tendency corresponding to the text information, and determine the public opinion risk level corresponding to the language information when it is determined that there is a public opinion risk;

[0015] The public opinion warning and processing module is used to generate warning information for prompting the anchor based on the language information when the public opinion risk level corresponding to the language information is higher than a preset threshold.

[0016] At least one embodiment of the present application also provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned method for real-time monitoring of public opinion risks in a live broadcast room.

[0017] At least one embodiment of the present application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-mentioned method for real-time monitoring of public opinion risks in a live broadcast room.

[0018] The real-time monitoring method, device, equipment and storage medium for public opinion risk in a live broadcast room provided by the embodiments of the present application collect the host's speech information and convert it into text information, thereby using a natural language model to perform context analysis and identification on the text information, so as to facilitate accurate identification of the semantics and emotional tendency of the host's sentences, and then accurately determine whether the collected data has public opinion risks, and strengthen the identification of texts that are not public opinion risk keywords but have potential negative effects, thereby improving the accuracy of public opinion risk identification. At the same time, a level assessment is conducted for the public opinion risks that have emerged, so as to facilitate the refined management of public opinion risks, and provide corresponding treatment measures for different public opinion risks, so as to facilitate timely prompting of the host to make corrections, improve the efficiency of resolving public opinion risks, and realize the intelligence and refinement of public opinion monitoring and response.

[0019] In some optional embodiments, the method further includes:

[0020] Collecting public opinion-related information of the live broadcast room; the public opinion-related information includes at least one of interactive information in the comment area of ​​the live broadcast room, public opinion information of the live broadcast enterprise, and evaluation information of the live broadcast product;

[0021] The determining whether the language information has a public opinion risk according to the semantic content and / or emotional tendency corresponding to the text information includes:

[0022] The semantic content and / or emotional tendency corresponding to the text information is combined with the public opinion-related information to determine whether the language information has a public opinion risk.

[0023] In some optional embodiments, the performing context analysis on the text information using a pre-built natural language model includes:

[0024] Using a pre-built keyword library to perform keyword recognition on the text information, and determining a first text containing a banned word or a sensitive word in the text information;

[0025] Filtering the text information according to the first text, and extracting a second text that has a public opinion risk in addition to the first text from the text information;

[0026] The second text is subjected to context analysis using a pre-built natural language model.

[0027] In some optional embodiments, the using a pre-built natural language model to perform context analysis on the text information to determine the semantic content and / or sentiment tendency corresponding to the text information includes:

[0028] Collecting sentence information associated with the text information; the sentence information includes sentences or phrases in the context adjacent to the text information;

[0029] Using a pre-built natural language model in combination with the sentence information, sequentially performing vocabulary analysis, grammatical structure analysis, and intonation analysis on the text information to determine the semantic content corresponding to the text information;

[0030] A pre-built natural language model is used in combination with the sentence information to perform contextual sentiment analysis on the text information to determine the sentiment tendency corresponding to the text information.

[0031] In some optional embodiments, judging whether the language information has a public opinion risk according to the semantic content and / or emotional tendency corresponding to the text information includes:

[0032] Matching the semantic content corresponding to the text information with the preset public opinion risk rules, determining a matching result, and judging whether the voice content has a public opinion risk based on the matching result;

[0033] The emotional information of the host corresponding to the text information is analyzed according to the emotional tendency corresponding to the text information, and based on the emotional information, it is determined whether the language information corresponding to the host has a public opinion risk.

[0034] In some optional embodiments, the public opinion risk rule includes at least one of a legal public opinion risk rule, a brand public opinion risk rule, and a live broadcast error public opinion risk rule.

[0035] In some optional embodiments, determining the public opinion risk level corresponding to the language information includes:

[0036] Determine the public opinion risk level and public opinion risk type corresponding to the voice information;

[0037] When the public opinion risk level corresponding to the language information is higher than a preset threshold, generating warning information for prompting the anchor based on the language information includes:

[0038] When the public opinion risk level corresponding to the language information is higher than a preset threshold, a warning message for prompting the anchor and a response strategy corresponding to the public opinion risk type are generated based on the public opinion risk type corresponding to the voice information. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] One or more embodiments are exemplarily described by the pictures in the corresponding drawings, and these exemplary descriptions do not constitute limitations on the embodiments.

[0040] Figure 1 This is a process of a method for real-time monitoring of public opinion risks in a live broadcast room provided by an embodiment of the present application. Figure 1 ;

[0041] Figure 2 This is a process of a method for real-time monitoring of public opinion risks in a live broadcast room provided by an embodiment of the present application. Figure 2 ;

[0042] Figure 3 This is a process of a method for real-time monitoring of public opinion risks in a live broadcast room provided by an embodiment of the present application. Figure 3 ;

[0043] Figure 4 This is a process of a method for real-time monitoring of public opinion risks in a live broadcast room provided by an embodiment of the present application. Figure 4 ;

[0044] Figure 5 is a schematic diagram of a device for real-time monitoring of public opinion risks in a live broadcast room provided by another embodiment of the present application;

[0045] Figure 6 is a structural schematic diagram of an electronic device provided by another embodiment of the present application;

[0046] Figure 7 It is a schematic diagram of a computer-readable storage medium provided by another embodiment of the present application. DETAILED DESCRIPTION

[0047] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the embodiments of the present application will be described in detail below in conjunction with the accompanying drawings. However, it will be appreciated by those skilled in the art that in the present application, many technical details are proposed in order to enable the reader to better understand the present application. However, even without these technical details and various changes and modifications based on the following embodiments, the technical scheme claimed in the present application can also be implemented. The division of the following embodiments is for the convenience of description, and the specific implementation of the present application should not be construed as any limitation, and the various embodiments can be combined and referenced with each other under the premise of no contradiction.

[0048] In order to facilitate understanding of the embodiments of the present application, relevant content about live broadcast public opinion monitoring is first introduced here.

[0049] With the rapid development of Internet technology, the live broadcast industry has ushered in unprecedented development. As an important platform for brand and user interaction, the compliance and positivity of the content in the live broadcast room are crucial to the shaping of the brand image. Therefore, it is particularly important to develop an efficient and accurate real-time monitoring system for public opinion risks in live broadcast rooms.

[0050] In the e-commerce live broadcast scenario, the real-time interaction between the anchor and the audience has become a key link in brand communication and product sales. However, the current public opinion monitoring system mainly relies on keyword matching technology, which is difficult to effectively identify complex language tendencies in the context, resulting in inaccurate identification of public opinion risks and untimely processing.

[0051] In order to solve the above-mentioned technical problem of inaccurate identification of public opinion risks, the present invention proposes a real-time monitoring method for public opinion risks in a live broadcast room. The implementation details of the real-time monitoring method for public opinion risks in a live broadcast room of this embodiment are specifically described below. The following content is only the implementation details provided for the convenience of understanding and is not necessary for the implementation of this solution.

[0052] Embodiment 1:

[0053] The method for real-time monitoring of public opinion risks in live broadcast rooms of this embodiment can be applied to electronic devices with communication, computing and data storage capabilities. The specific process can be as follows: Figure 1 As shown, steps 101-105 are included, specifically including:

[0054] Step 101, collecting language information of the live broadcast room; the language information includes the host's speech information.

[0055] Specifically, when the anchor in the live broadcast room speaks, the system will capture the language information in real time, and then monitor the current live broadcast room in real time. The language information can also include the interactive information and / or barrage information in the comment area of ​​the live broadcast room, so as to monitor public opinion through the interactive information and / or barrage information in the comment area. In this embodiment, after the language information is collected, the language information is clarified, and irrelevant characters, punctuation marks, labels, etc. are removed to ensure the purity of the data.

[0056] In some examples, the live broadcast room is a live broadcast room for shopping. When collecting language information, public opinion-related information of the live broadcast room can be collected simultaneously; the public opinion-related information includes at least one of the interactive information in the comment area of ​​the live broadcast room, the public opinion information of the live broadcast enterprise, and the evaluation information of the live broadcast product, so as to facilitate public opinion risk monitoring in combination with the public opinion-related information and improve monitoring accuracy.

[0057] Among them, the live broadcast enterprise public opinion information includes the current social public opinion information of the enterprise that opened the live broadcast room and the current social public opinion information of the enterprise that sells products through live broadcast, so as to facilitate the monitoring of the live broadcast room and avoid the risk of adverse public opinion.

[0058] Step 102: Identify the language information and determine text information corresponding to the language information.

[0059] Specifically, the collected language information can be processed into text information through relevant tools to facilitate subsequent analysis and processing. The relevant tools can use speech recognition technology to convert the language information of the anchor's speech information into text information.

[0060] Furthermore, the system also performs pre-processing work such as word segmentation and part-of-speech tagging on text information to provide a basis for subsequent analysis. Specifically, the data is cleaned to remove irrelevant characters, punctuation, labels, etc. to ensure the purity of the data; the data is segmented to divide the text into independent words or sub-words for better understanding by the model; the data is annotated to mark public opinion risk points, such as sensitive words, negative comments, etc., for model training and supervised learning.

[0061] Step 103: Use a pre-built natural language model to perform context analysis on the text information to determine the semantic content and / or sentiment tendency corresponding to the text information.

[0062] Specifically, when constructing a natural language model, the contextual information of the target text is used for association training, so that the natural language model can connect the context and infer the emotional tendency through the contextual context. When analyzing the context through the natural language model, the contextual context of the language information corresponding to the text information can be combined for analysis, thereby accurately inferring the semantic content and / or emotional tendency corresponding to the text information, thereby improving the accuracy of public opinion risk monitoring.

[0063] In some embodiments, step 103, such as Figure 2 As shown, including:

[0064] 1031. Perform keyword recognition on the text information using a pre-built keyword library to determine the first text containing a banned word or a sensitive word in the text information;

[0065] 1032. Screen the text information according to the first text, and extract a second text that has a public opinion risk in addition to the first text from the text information;

[0066] 1033. Perform context analysis on the second text using a pre-built natural language model.

[0067] Specifically, by first performing keyword matching to screen out the first text containing banned words or sensitive words, and then combining the natural language model to perform contextual analysis on the second text in the text information that has public opinion risks other than the first text, it is possible to combine multiple methods to monitor public opinion risks and improve the accuracy of public opinion risk monitoring.

[0068] Among them, the second text may include all text information except the first text, and may also include text information that has public opinion risks other than the first text. After the first text is screened out in the system, the remaining text information can be expanded and identified through a natural language model to identify synonyms and antonyms related to banned words or sensitive words, thereby facilitating the screening of the second text that has public opinion risks other than the first text, thereby further performing contextual analysis on the second text to improve the accuracy of public opinion monitoring.

[0069] In some embodiments, step 103, such as Figure 3 As shown, including:

[0070] Step 201: Collect sentence information associated with the text information; the sentence information includes sentences or phrases in the context adjacent to the text information;

[0071] Step 202: using a pre-built natural language model in combination with the sentence information, sequentially performing vocabulary analysis, grammatical structure analysis, and intonation analysis on the text information to determine the semantic content corresponding to the text information;

[0072] Step 203: Use a pre-built natural language model in combination with the sentence information to perform contextual sentiment analysis on the text information to determine the sentiment tendency corresponding to the text information.

[0073] Specifically, by collecting sentence information, the natural language model can combine the context of the text information for association identification and accurately infer the emotional tendency corresponding to the text information, so as to improve the accuracy of monitoring public opinion.

[0074] In this embodiment, in step 202, vocabulary analysis includes: the model will identify emotional words in the text, such as positive words (such as "happy", "like") and negative words (such as "disappointed", "hate"). 2. Grammatical structure analysis includes: the model will analyze the grammatical structure of the text, such as the type of sentences (declarative sentences, interrogative sentences, exclamatory sentences, etc.) and the logical relationship between sentences. 3. Intonation analysis includes: although the intonation information cannot be directly obtained in the context of text form, the model can infer the intonation through punctuation (such as exclamation marks, question marks, etc.) and the overall style of the text. Through multiple analysis steps, the text information can accurately infer the corresponding emotional tendency through the natural language model, thereby improving the accuracy of public opinion monitoring.

[0075] Among them, when collecting sentence information associated with the text information, the global context and local context information of the text information can be collected. Global context: refers to the structure of the entire discourse or text, examining the position of the current sentence in the hierarchy of the conversation or text. For example, in a complete conversation or article, the model needs to understand how the various parts are related to each other to form a whole meaning. 2. Local context: refers to the sentences or phrases directly adjacent to the current sentence, examining the connection between the current sentence and these sentences or phrases.

[0076] By accurately collecting sentence information, the natural language model can identify coherent semantic content by understanding the collocation information of previous and subsequent words or phrases, thereby improving the accuracy of public opinion risk monitoring.

[0077] Step 104, judging whether the language information has a public opinion risk based on the semantic content and / or emotional tendency corresponding to the text information, and determining the public opinion risk level corresponding to the language information when it is determined that there is a public opinion risk.

[0078] Specifically, by analyzing the language information that reveals public opinion risks and classifying them into public opinion risk levels, it is convenient to carry out refined management of public opinion monitoring, so as to provide corresponding treatment measures according to the public opinion risk levels and accurately resolve public opinion risks.

[0079] In some embodiments, step 104 includes using the semantic content and / or emotional tendency corresponding to the text information in combination with the public opinion-related information to determine whether the language information has a public opinion risk.

[0080] Specifically, by combining text information with public opinion-related information, not only the anchor's language content is taken into account, but also the interactive information in the comment area of ​​the live broadcast room, the public opinion information of the live broadcast enterprise, and the evaluation information of the live broadcast product, so that the monitoring of public opinion risks can be comprehensively identified from multiple angles, and whether the language information has public opinion risks can be accurately identified, so as to improve the accuracy of risk monitoring. In this embodiment, the live broadcast enterprise refers to the brand side of the live broadcast product.

[0081] In some embodiments, it is determined whether the language information has a public opinion risk based on the semantic content and / or emotional tendency corresponding to the text information, such as Figure 4 As shown, including:

[0082] Step 301: Match the semantic content corresponding to the text information with a preset public opinion risk rule, determine a matching result, and judge whether the semantic content has a public opinion risk based on the matching result;

[0083] Step 302: Analyze the emotional information of the host corresponding to the text information according to the emotional tendency corresponding to the text information, and determine whether the language information corresponding to the host has a public opinion risk based on the emotional information.

[0084] Specifically, by analyzing the matching results of public opinion risk rules and the emotional information of the anchor, and combining the analysis of whether the language information poses a public opinion risk, it is possible to provide a variety of public opinion risk monitoring content, thereby improving the accuracy of public opinion risk monitoring.

[0085] In some examples, the public opinion risk rules include at least one of the legal public opinion risk rules, brand public opinion risk rules, and live broadcast error public opinion risk rules. The legal public opinion risk rules are preset with relevant illegal words or short sentences. Based on the fact that the live broadcast room in this embodiment is a live broadcast room for selling goods, the legal public opinion risk rules usually preset relevant keywords based on the Advertising Law. The brand public opinion risk rules are risk rules formulated based on relevant remarks that harm the brands selling goods; the live broadcast error public opinion risk rules are risk rules formulated based on the host's remarks introducing errors.

[0086] Furthermore, corresponding public opinion risk types are set corresponding to different public opinion risk rules, so as to determine the public opinion risk level corresponding to the language information, including: determining the public opinion risk level and public opinion risk type corresponding to the voice information.

[0087] By classifying and processing the types of public opinion risks, we can provide targeted and corresponding treatment measures when public opinion risks exist, facilitate accurate response to public opinion risks, and resolve risks in a timely manner.

[0088] Step 105: When the public opinion risk level corresponding to the language information is higher than a preset threshold, a warning message for prompting the anchor is generated based on the language information.

[0089] Specifically, the preset threshold is the public opinion risk level threshold. After evaluating the public opinion risk level of the language information, the severity of the public opinion risk can be analyzed in combination with the preset threshold, and then corresponding warning processing can be provided to facilitate the anchor to correct the content in time and avoid public opinion risks.

[0090] In some embodiments, step 105 includes, when the public opinion risk level corresponding to the language information is higher than a preset threshold, generating a warning message for prompting the host based on the public opinion risk type corresponding to the voice information and a response strategy corresponding to the public opinion risk type.

[0091] Specifically, the system pre-builds response strategies corresponding to different types of public opinion risks, so that when it is analyzed that there is a public opinion risk in the current live broadcast room, corresponding response strategies can be provided to avoid the identified public opinion risk type, thereby achieving effective control of public opinion risks.

[0092] Among them, the public opinion risk types correspond to different public opinion risk rules, such as legal public opinion risk type, brand public opinion risk type, and live broadcast error public opinion risk type, so as to facilitate targeted processing and improve the efficiency and effect of public opinion risk control. The response strategies may include generating correction words corresponding to the public opinion risk type for the anchor to explain and correct, reporting to the manager, generating an apology video for live broadcast, etc.

[0093] In some examples, the public opinion risk level can be divided into different levels such as minor, medium, and severe. Examples of pre-built response strategies are as follows: For example, 1. The system combines the host's speech and the parameters of the products on sale in the mall to determine that the host's slip of the tongue has led to errors in the introduction of product specifications. At the same time, it detects evidence in the comment interaction. The system determines that a minor public opinion crisis has occurred and performs relevant processing, such as on-site prompts or generating subtitles to correct the host. This level of public opinion crisis has a small impact, so simpler response measures are adopted to avoid interrupting the live broadcast rhythm; 2. If the host says words such as "biggest" and "strongest", although they are not banned words, the system combines the context and semantic tendency to determine that this violates the Advertising Law, and can The system can cause the company to be fined and judge this situation as a moderate public opinion crisis that may result in financial losses. The anchor will be immediately prompted to correct the situation, and relevant apology words and legal science knowledge will be generated. The anchor will apologize and briefly popularize some knowledge of advertising law to solve the public opinion crisis in time. 3. If the anchor's remarks contain strong negative emotions, offense, sarcasm, etc., the system will judge that this public opinion crisis may cause huge losses to the company's brand image, and will generate prompts and relevant apology statements in time, and immediately contact the director's room, and the anchor's superior will come forward to explain and apologize, so as to deal with the public opinion in a controllable range in time.

[0094] The real-time monitoring method for public opinion risks in live broadcast rooms provided in this embodiment collects the host's speech information and converts it into text information, thereby using a natural language model to perform contextual analysis and identification on the text information, so as to accurately identify the semantics and emotional tendencies of the host's sentences, and then accurately determine whether the collected data has public opinion risks, and strengthen the recognition of texts that are not originally public opinion risk keywords but have potential negative effects, thereby improving the accuracy of public opinion risk identification. At the same time, a level assessment is performed on the public opinion risks that have emerged, so as to facilitate the refined management of public opinion risks, and provide corresponding treatment measures for different public opinion risks, so as to facilitate timely prompting of the host to make corrections, improve the efficiency of resolving public opinion risks, and realize the intelligence and refinement of public opinion monitoring and response. This embodiment can realize the accurate identification and intelligent response to public opinion risks in e-commerce live broadcasts, and improve the brand's public opinion management capabilities in live broadcast scenarios.

[0095] Embodiment 2:

[0096] Another embodiment of the present application relates to a real-time monitoring device for public opinion risks in a live broadcast room. The implementation details of the real-time monitoring device for public opinion risks in a live broadcast room of this embodiment are specifically described below. The following content is only the implementation details provided for the convenience of understanding and is not necessary for the implementation of this solution. The schematic diagram of the real-time monitoring device for public opinion risks in a live broadcast room of this embodiment can be as follows Figure 5 As shown, it includes: a data collection module 801, a text extraction module 802, an analysis module 803, a public opinion risk judgment module 804 and a public opinion early warning and processing module 805.

[0097] The data collection module 801 is used to collect language information of the live broadcast room; the language information includes the host's speech information;

[0098] A text extraction module 802, used to identify the language information and determine the text information corresponding to the language information;

[0099] An analysis module 803 is used to perform context analysis on the text information using a pre-built natural language model to determine the semantic content and / or sentiment tendency corresponding to the text information;

[0100] The public opinion risk judgment module 804 is used to judge whether the language information has a public opinion risk according to the semantic content and / or emotional tendency corresponding to the text information, and determine the public opinion risk level corresponding to the language information when it is determined that there is a public opinion risk;

[0101] The public opinion warning and processing module 805 is used to generate warning information for prompting the anchor based on the language information when the public opinion risk level corresponding to the language information is higher than a preset threshold.

[0102] It is worth mentioning that all modules involved in this embodiment are logic modules. In practical applications, a logic unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. In addition, in order to highlight the innovative part of this application, this embodiment does not introduce units that are not closely related to solving the technical problems proposed by this application, but this does not mean that there are no other units in this embodiment.

[0103] Embodiment three:

[0104] Another embodiment of the present application relates to an electronic device, such as Figure 6 As shown, it includes: at least one processor 901; and a memory 902 that is communicatively connected to the at least one processor 901; wherein the memory 902 stores instructions that can be executed by the at least one processor 901, and the instructions are executed by the at least one processor 901 so that the at least one processor 901 can execute the real-time monitoring method for public opinion risks in the live broadcast room in the above-mentioned embodiments.

[0105] Among them, the memory and the processor are connected in a bus manner, and the bus may include any number of interconnected buses and bridges, and the bus connects various circuits of one or more processors and memories together. The bus can also connect various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be one element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices on a transmission medium. The data processed by the processor is transmitted on a wireless medium via an antenna, and further, the antenna also receives data and transmits the data to the processor.

[0106] The processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory can be used to store data used by the processor when performing operations.

[0107] Embodiment 4:

[0108] Another embodiment of the present application relates to a computer-readable storage medium, such as Figure 7 As shown, a computer program 31 is stored. When the computer program 31 is executed by a processor, the above method embodiment is implemented.

[0109] That is, those skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium, including a number of instructions to enable a device (which can be a single-chip microcomputer, chip, etc.) or a processor to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as: ROM), random access memory (Random Access Memory, referred to as: RAM), disk or optical disk and other media that can store program codes.

[0110] Those skilled in the art will appreciate that the above embodiments are specific embodiments for implementing the present application, and in actual applications, various changes may be made thereto in form and detail without departing from the spirit and scope of the present application.

Claims

1. A method for real-time monitoring of public opinion risks in a live broadcast room, characterized in that: include: Collecting language information of the live broadcast room; the language information includes the host's speech information; Identifying the language information and determining text information corresponding to the language information; Performing context analysis on the text information using a pre-built natural language model to determine the semantic content and / or sentiment tendency corresponding to the text information; Determine whether the language information has a public opinion risk based on the semantic content and / or emotional tendency corresponding to the text information, and determine the public opinion risk level corresponding to the language information when it is determined that there is a public opinion risk; When the public opinion risk level corresponding to the language information is higher than a preset threshold, a warning message for prompting the anchor is generated based on the language information.

2. According to claim 1, a method for real-time monitoring of public opinion risks in a live broadcast room is characterized in that: The method further comprises: Collecting public opinion-related information of the live broadcast room; the public opinion-related information includes at least one of interactive information in the comment area of ​​the live broadcast room, public opinion information of the live broadcast enterprise, and evaluation information of the live broadcast product; The determining whether the language information has a public opinion risk according to the semantic content and / or emotional tendency corresponding to the text information includes: The semantic content and / or emotional tendency corresponding to the text information is combined with the public opinion-related information to determine whether the language information has a public opinion risk.

3. According to claim 1, a method for real-time monitoring of public opinion risks in a live broadcast room is characterized in that: The using a pre-built natural language model to perform context analysis on the text information includes: Using a pre-built keyword library to perform keyword recognition on the text information, and determining a first text containing a banned word or a sensitive word in the text information; Filtering the text information according to the first text, and extracting a second text that has a public opinion risk in addition to the first text from the text information; The second text is subjected to context analysis using a pre-built natural language model.

4. A method for real-time monitoring of public opinion risks in a live broadcast room according to claim 1 or 3, characterized in that: The using of the pre-built natural language model to perform context analysis on the text information to determine the semantic content and / or sentiment tendency corresponding to the text information includes: Collecting sentence information associated with the text information; the sentence information includes sentences or phrases in the context adjacent to the text information; Using a pre-built natural language model in combination with the sentence information, sequentially performing vocabulary analysis, grammatical structure analysis, and intonation analysis on the text information to determine the semantic content corresponding to the text information; A pre-built natural language model is used in combination with the sentence information to perform contextual sentiment analysis on the text information to determine the sentiment tendency corresponding to the text information.

5. According to claim 1, a method for real-time monitoring of public opinion risks in a live broadcast room is characterized in that: The determining whether the language information has a public opinion risk according to the semantic content and / or emotional tendency corresponding to the text information includes: Matching the semantic content corresponding to the text information with the preset public opinion risk rules, determining a matching result, and judging whether the semantic content has a public opinion risk based on the matching result; The emotional information of the host corresponding to the text information is analyzed according to the emotional tendency corresponding to the text information, and based on the emotional information, it is determined whether the language information corresponding to the host has a public opinion risk.

6. A method for real-time monitoring of public opinion risks in a live broadcast room according to claim 5, characterized in that: The public opinion risk rules include at least one of legal public opinion risk rules, brand public opinion risk rules and live broadcast error public opinion risk rules.

7. A method for real-time monitoring of public opinion risks in a live broadcast room according to claim 1, characterized in that: Determining the public opinion risk level corresponding to the language information includes: Determine the public opinion risk level and public opinion risk type corresponding to the voice information; When the public opinion risk level corresponding to the language information is higher than a preset threshold, generating warning information for prompting the anchor based on the language information includes: When the public opinion risk level corresponding to the language information is higher than a preset threshold, a warning message for prompting the anchor and a response strategy corresponding to the public opinion risk type are generated based on the public opinion risk type corresponding to the voice information.

8. A real-time monitoring device for public opinion risk in a live broadcast room, characterized in that: include: A data collection module is used to collect language information of the live broadcast room; the language information includes the host's speech information; A text extraction module, used to identify the language information and determine the text information corresponding to the language information; An analysis module, used to perform context analysis on the text information using a pre-built natural language model to determine the semantic content and / or sentiment tendency corresponding to the text information; A public opinion risk judgment module is used to judge whether the language information has a public opinion risk according to the semantic content and / or emotional tendency corresponding to the text information, and determine the public opinion risk level corresponding to the language information when it is determined that there is a public opinion risk; The public opinion warning and processing module is used to generate warning information for prompting the anchor based on the language information when the public opinion risk level corresponding to the language information is higher than a preset threshold.

9. An electronic device, characterized in that: include: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the real-time monitoring method for public opinion risks in a live broadcast room as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for real-time monitoring of public opinion risks in a live broadcast room described in any one of claims 1 to 7 is implemented.

Images